Measuring country level partisanship with ESS data a new approach

Transcription

1 Measuring country level partisanship with ESS data a new approach Veronika PATKÓS, Hungarian Academy of Sciences, Centre for Social Sciences, Institute for Political Science; Corvinus University of Budapest Presented in session: Political attitudes and voting at the 3rd International ESS Conference, 13 15th July 2016, Lausanne, Switzerland Abstract The starting point of this paper is that conventional methods to measure partisanship do not reflect an important element of partisanship that scholars generally agree on. This element is that partisanship is a cognitive mechanism, a perceptual screen that helps information processing and helps to assess what happens on the political scene. The method presented here tries to include this theoretically important feature of partisanship, by using the first six rounds of ESS data. Its scope is to integrate positive and negative biases that voters have toward political sides in one variable. It relies on the relative difference between the overall political evaluations of voters of governing and opposition parties. Across this difference the new indicator shows how strong is the partisan polarisation of voters perception in a country. It offers a completely different picture compared to the existing rankings, actually, it is negatively correlated to partisanship measured with the closeness-to-a-party question. The level of partisan polarisation in a country is associated with lower levels of press freedom, with higher level of ideological polarisation, with a more majoritarian political system, and with lower levels of institutional trust in the electorate, while there is no relationship between partisan polarisation and the length of democratic experience. This implies that the usability of conventional methods to take inferences on how much partisan bias is present in voters thinking is highly questionable. Keywords Partisanship, partisan bias, voting behaviour, Europe Introduction But can we infer that a single response to a closed-ended question necessarily involves a deep and meaningful political commitment? (Klingemann & Wattenberg, 1992, p. 132) Several recent studies investigating voting behavior defined partisanship as a cue in political information processing, which helps voters to harmonize and integrate new information with existing beliefs. In this vein, partisanship leads to a biased opinion formation, a somewhat prejudiced or blind attitude toward new political information, mainly if new information is dissonant to existing attitudes. While many scholars focus on this aspect of partisanship, the most widely used measurement methods are still based upon electoral volatility, formal institutional linkages to parties, or self-reported, closeness-to-a-party type questions. The methodological problem with these kinds of measurements is that such indicators tell little or nothing about the extent of partisan bias in voters opinion formation, although that is an essential part of the most widely used definitions. 1

2 This paper aims at elaborating a new method for measuring partisanship with ESS data, which can be used for the purpose of country-level comparisons. This method relies on the differences between how voters of opposition parties and voters of governing parties view and evaluate political institutions. It uses the first six rounds of ESS data about institutional trust and evaluation of the functioning of the political system. The aim of this methodological innovation is to create a new method for measuring partisanship, which serves better for an international comparison and is more in line with the above presented, bias-generating feature of the concept. The presented aspect of the concept may be called partisan polarisation. While partisanship and polarisation are two distinct concept analytically, they are empirically closely related to each other. Describing the functioning of the polarised Hungarian political system, Palonen points that parties or camps exist through their common opposition to one another, with a consequent normative-ideological logic: as you are the bad ones, we are the good ones. This description is very similar to how Iyengar, Sood and Lelkes write about partisanship in the US: the sense of partisan identity is increasingly associated with a Manichean, us against them view of the political world. Democrats and Republicans harbor generally negative feelings toward their opponents ( ) there is sufficient animosity to make partisan affiliation relevant to inter-personal relations (Iyengar, Sood, & Lelkes, 2012, p. 421). Based on this, we may declare a country to be strongly polarised in a partisan sense if partisan attachments are an influential predictor of how voters evaluate political information, and if its political scene is dominated by a my party right or wrong (Klingemann & Wattenberg, 1992, p. 131) attitude. The country-level results obtained with this new method present a rather different pattern compared to the existing rankings, they are (weakly, but significantly) negatively correlated with the results based on the proportion of those who felt themselves closer to one party than to the others. Moreover, measured this way partisanship is negatively correlated with important factors as trust in political institutions, the number of effective parties and with the level of fractionalization of the party system, political-economic stability, or economic performance of the country. This implies that formerly used methods focus on a certain aspect of partisanship, but are immune to another and very important aspect of the phenomenon. Based on these results it is doubtful whether strong partisan feelings have a positive effect on democratic functioning, which is a general assumption in the literature. Concept Party identification and partisanship are central concepts of the study of voting behaviour since Campbell and his co-authors wrote their seminal work (Campbell, Converse, Miller, & Stokes, 1960). These concepts have their roots in the works of these authors, and the literature still relies strongly on the definitions presented there. However, despite the existence of a well-known common root, definitions may differ considerably from each other across works and authors. The main cause of this variety is that even the original work offers multiple definitions, which contain a range of features rather than some clearly defined elements. Therefore, while scholars agree that partisanship is an underlying attachment to a party, they choose different supplementary elements from the original definitions when describing the concept. Table 1 shows a sorting of how various works approach partisanship by checking for some key elements which regularly appear in the literature since As we can see, some authors stress mainly the longtermness of the attachment, many others view it as a heuristic in decision making, or as a perceptual screen which helps to integrate and evaluate new political information. A smaller number of researchers define partisanship as a special 2

3 group-attachment, and very few of them think self-definition to be indispensable when identifying who is a partisan voter. On the whole, an overwhelming majority of scholars agree that partisanship is a longterm attachment to a party, which has an important role on how individuals process political information and how they make political decisions. It is important to note that among the substantial features of partisanship the option perceptual screen, a cognitive mechanism which helps voters to process new political information is the most widely mentioned 1. Table 1. Features of partisanship in the literature. Psychological attachment to a party Stress on the importance of self-definition Stress on the longtermness of the attachment Heuristic, cost-saving device for voters Perceptual screen, a cognitive mechanism which structures political information Group attachment (Campbell et al., 1960) (Converse & Pierce, 1985) Jacoby 1988 (Fiorina, 1990) (Green & Palmquist, 1994) (Rose & Mishler, 1998) (Miller & Klobucar, 2000) (Bartels, 2002) (Green et al., 2004) (Tóka, 2005) (Dalton & Weldon, 2007) (Shapiro & Bloch-Elkon, 2008) (Gerber & Huber, 2009) (Budge et al., 2010) (Tilley & Hobolt, 2011) (Anduiza et al., 2013) (Lupu, 2013) (Achen & Bartels, 2016) To pick some examples, partisanship is a heuristic that helps individuals to organize the complexities of politics, integrate information into a political belief system and evaluate political phenomena (Dalton & Weldon, 2007, pp ); it influences voters policy stances, their evaluation of the economy, attitudes toward democracy, and electoral behavior (Samuels, 2006); and it structure[s] one's cognitive understanding of politics (Miller & Klobucar, 2000, p. 675). Partisanship shapes the way citizens see the world of politics and public affairs to an extent that they practically tend to see what they want to see (Gerber & Huber, 2009, p. 423). 1 Clearly, another selection of sources might offer a different picture, while I don t think that it would radically change the above presented. 3

4 From this point of view it is striking how the most widely used measurement methods ignore such an important aspect of partisanship. The traditional questions 2 which are used to measure partisanship rely strongly on self-definition and on the willingness of subjects to report their partisan attachments, while the idea of using volatility to measure country-level partisanship relies on the longtermness of the attachments. These methods do not have any element that refers to information processing, which indicates that they are not very useful for taking inferences on this point, even if many scholars use them for this purpose. Data and method In order to include this conceptually important feature of partisanship in the measurement method, I propose a new way for measuring the extent of partisan bias in information processing on country level, or, to put it in another way, the level of partisan polarisation of a country. This method relies on the differences between the political opinions of the competing political camps. Comparing (but even identifying) these political camps in different party systems which are composed of various number of parties of various size and ideological position is quite challenging. An option to make a distinction between political camps is to use parties winner-loser status, based on whether they were governing parties or opposition parties during the fieldwork period of ESS surveys. At first sight, identifying the two camps in every countries based on their winner-loser status may seem a major simplification. But in my opinion it is still the least questionable way to create comparable groups by all countries electorate, as it fundamentally mirrors who are with who on the political scene of the given country. Comparing the political evaluation of voters based on their winner-loser status is not a novelty, there is a growing body of literature that approaches the so called winner-loser gap (Anderson & LoTempio, 2002; Anderson & Tverdova, 2001; Blais & Gélineau, 2007; Brunell & Buchler, 2012; Craig, Martinez, Gainous, & Kane, 2006; Curini, Jou, & Memoli, 2012; Howell & Justwan, 2013; Singh, Karakoç, & Blais, 2012; Singh, Lago, & Blais, 2011). In contrast to this growing scholarly attention, it is interesting that the winner-loser gap has not yet been explicitly linked to partisan polarisation. Therefore, I propose to create winner-loser groups in each country, and to make an index by subjects answers about the trust and evaluation of different dimensions of the political system (including trust in parliament, trust in the police, trust in the legal system and trust in politicians; satisfaction with how democracy works, satisfaction with the national government, satisfaction with the economy and satisfaction with the educational and healthcare systems). I decided to use all these variables about institutional trust and political satisfaction in order to offer a more complete picture about voters political views and evaluations, and to mitigate the effect of eventual country-specific associations between certain variables. Instead of the index, another option could be to create a factor. From a practical point of view there is a negligible difference between the index and the factor, since there is an extremely high correlation between them (r=0,999). To yield more easily interpretable results I decided to use the index. For my analysis I used the first six rounds of ESS data, excluding countries which are not parts of the EU or of the Schengen area. In order to make a clear distinction between winners and losers, I also excluded countries in those cases when fieldwork period overlapped with 2 In the European political context the standard questions (used also by the ESS) referring to partisanship are Do you feel closer to one particular party than all the others? How close do you feel to this party? Which party do you feel closer to? In the political context of the US the standard question is Generally speaking, do you think (or sometimes: feel) to be a Democrat, a Republican, an Independent, or what? 4

5 parliamentary elections, when surveys were conducted after the parliamentary elections, and when the new government has not yet entered in office. In two cases I excluded the voters of those parties which left the government in fieldwork period. Neither do I include cases when the incumbent government was a technocratic one. All these exceptions are listed on Table 2. The process finally resulted in 122 subjects on the level of countries (see Appendix 1), relying on a database of individual subjects from 30 countries. Data about government composition, political institutional and economic variables have been borrowed from the Comparative Political Data Set project (Armingeon et al. 2013; 2015). In order to avoid reverse ecological fallacy, making an index by adding more variables is only justified if they are positively correlated in each countries in all (or almost all) of the countries (Hofstede, 1984). The nine variables of institutional trust and political satisfaction are positively correlated in the overwhelming majority of cases: we find a significant positive relationship (p<0,01) between them in 4378 cases of The only exception is Slovakia in the 2 nd ESS round; where there is no significant correlation between some of the satisfaction and trust variables. Based on this feature of the data I found it legitimate to make an index of the nine variables. Table 2. Countries excluded from the analysis. Parliamentary elections during fieldwork period Parliamentary elections prior to the fieldwork period, new government not yet entered in office Voters of parties who left the government during fieldwork period Technocratic government in office Countries excluded from the analysis Number of countries in the sample Estonia, Croatia, Germany, Italy, Netherla Ireland, Greece Cyprus, Sweden Portugal nds Netherlands Estonia LPF voters (Netherlands) Slovenia Czech Republic Netherlands NC voters (France) Albania, Kosovo, Russia, Turkey, Ukraine, Israel Instead of counting the differences between the mean evaluations of winners and losers I think it is better to divide the mean of winners ratings by the mean of losers. To take an example, in 2012 the net difference between winners and losers ratings was 0.46 in Bulgaria and 0.41 in Finland on a 0 to 10 scale, so the net difference between winners and losers differ little, based on these values the level of partisan polarisation in Bulgaria and in Finland is very similar. But the averages of the evaluations in the two countries are 2.85 and 6.71 respectively, and this information sheds new light on the two countries level of partisan polarisation. Since I think this aspect is meaningful, I prefer to use the relative difference (dividing winners evaluation by the evaluation of losers) instead of net differences. Results Figure 1 shows the level of the created index by country and by ESS round. Apart from the results of individual countries that will be discussed later, there are two striking characteristics. One of them is that the typical level of partisan polarisation is between 1.05 and 1.2. In the vast majority of countries (in 116 cases out of 122) winners are more satisfied 5

6 and trusting than losers, but generally not to an exaggerated extent. ANOVA tests show that out of this 116 cases the difference between the two groups is significant in 108 cases, while in three cases we find a weak but significant negative relationship between political evaluations and winning position (in Finland, sample of 2003; Belgium, sample of 2007 and Slovenia, sample of 2009). The other important feature to be noted is the volatility of partisan polarisation within cases: however some countries results do not differ too much by ESS rounds, in other cases like Hungary, Greece, Spain, Lithuania, Bulgaria, Slovenia, Poland, and the Czech Republic there is considerable difference between them, despite of the relatively short time span. This feature shows that even if structural factors may also explain one part of the variance of partisan polarisation, there is a considerable part of its variance that may be caused by nonstructural factors, for example by an economic or political crisis, or by the changing of the dominant style of political communication in a country. Figure 1. The level of partisan polarisation (relative difference of winners and losers political evaluations) in European countries by ESS rounds. The reference line indicates where the evaluations of winners and losers are identical. Turning to the puzzling cases, two countries certainly deserve attention for outlier results. One is Slovenia with its extremely negative winners ratings compared to the losing camp; the 6

7 other is Hungary, where polarisation is by far the strongest in the analysed period. Such an extreme level of partisan polarisation is surprising, even if scholars recurrently report very strong partisan feelings in Hungarian society ((Körösényi, 2013; Palonen, 2009; Tardos & Angelusz, 2009; Tóka, 2005), in contrast to feeling-close-to-a-party type questions, which have never shed light on this feature of the electorate. These interesting cases are worthwhile to be studied in future works. To have an idea about what do these results tell us about a countries political system, I investigate the relationship between partisan polarisation and some country-level political characteristics. The new indicator of partisan polarisation and one of the most conventional indicators (the proportion of those voters who stated they are closer to a party than to the others) are very weakly, but significantly negatively correlated (r=0.158, p=0.030). The second step is to explore the relationship between some variables of the quality of democratic functioning and partisanship, measured both ways. A general assumption in the literature is that strong partisan feelings are indispensable for healthy democratic functioning, and that longer democratic experience is associated to stronger attachment to parties. Thus, the assumption is that at least one part of Eastern-European countries democratic problems comes from the fact that party attachments in these countries are generally weak (Converse, 1969; Klingemann & Wattenberg, 1992; Rose & Mishler, 1998a; Tóka, 2005). In order to explore these relationships I analysed correlations between country-level variables and indicators of the level of partisanship. Table 3. Correlations between partisanship (measured in two ways) and political characteristics of countries. Democratic experience Press freedom Ideological polarisation 7 Effective number of parties Consensual/ majoritarian democracy Institutio nal constrain ts Overall trust and satisfaction in political institutions Level of partisanship (traditional method) ** *** Level of partisan *** 0.401*** * polarisation (new *** *** *** method) N ***Correlation is significant at the 0.01 level (2-tailed). **Correlation is significant at the 0.05 level (2-tailed). * Correlation is significant at the 0.1 level (2-tailed). Standard errors are adjusted for 30 clusters (countries as clusters). Democratic experience indicates the number of years after 1945 spent with democratic institutions according to Polity IV. Press freedom indicates the value of freedom of the press according to Freedom House. Higher values mean press is less free. Ideological polarisation indicates how much are far left and right political camps from each other. It uses the lrscale variable of ESS rounds, and it shows the distance of median positions of those who put themselves on the left side on the scale (0-4) and those who put themselves on the right side of the scale (6-10). Higher values mean the electorate is more polarized ideologically. Effective number of parties: the source of this variable is the CPDS data base, the original name of the variable is effpart_leg. Counted based on the seats level according to the formula proposed by Laakso and Taagepera (1979). Consensual/majoritarian democracy: Lijphart s first dimension, it shows the level of how much a country is consensual. The source of the variable is the CPDS data base, the original name of the variable is lfirstp. Higher values indicate more consensual political functioning. Institutional constraints: High values indicate powerful constraints and less room for maneuverability for the central government. The index is composed by six dummy variables regarding EU membership, degree of

8 centralisation of state structure,difficulty of amending constitutions, strong bicameralism, central bank autonomy and frequent referenda. The source of the variable is the CPDS data base, the original name of the variable is instcons. Overall trust and satisfaction: it is the mean of scores about trust in parliament, in the legal system, in the police, in politicians; and satisfaction with democracy, with the national government, with the economy, and with the healthcare and educational systems, on a 0 to 10 scale. Results show that the traditional method produce an indicator which is less consistently associated with some important features of democratic functioning. We find only two significant correlations, which show that if partisanship is measured based on the willingness of voters to declare they are close to a party, partisanship is rather strongly associated to overall institutional trust and satisfaction with political functioning. The other significant correlation is that of between partisanship and freedom of the press higher freedom is associated with a higher proportion of people who state to be close to a party. In my opinion these results are plausible, but not very telling. We may argue that if partisanship is measured based on the willingness of reporting party attachments, these results show that more trustable political systems have more partisan voters. In contrast to two general assumptions, we do not find a significant relationship nor between partisanship and the length of one country s democratic experience, nor between partisanship and ideological polarisation. Turning to the new indicator, which relies on the perceptual screen feature of partisanship, we find relatively strong correlations, which show a consistent picture. 3 Measured this way, higher levels of partisanship are associated to a higher level of ideological polarisation in the electorate, to reduced levels of press freedom, to a more majoritarian political system, to a wider room to manoeuvre from the part of the central government, and to lower levels of electoral trust and satisfaction towards institutions. All in all, recalling the general assumptions which see partisanship as a guarantee of democratic functioning, the picture offered here is completely different. We may rather assume the exact contrary of this assumption, namely, that heightened partisan feelings in the electorate are associated with a less democratic functioning. At this point, this question should be left to be studied later. Conclusion The starting point of this paper was that the most conventional methods to measure the level of partisanship in a country (such as direct survey questions or electoral volatility) have a common weakness. They do not reflect the element of partisanship that scholars generally agree on: its feature of being a cognitive mechanism, a perceptual screen in information processing and in assessing what happens on the political scene. As Burden and Klofstad state upon assuming the Michigan theory of party identification as an affective social identity, the traditional means of measurement are something of a curiosity (2005, p. 880). Therefore, there is a theoretically widely accepted element of partisanship (we may call it partisan bias) which does not appear in the most widely used methods. Another weakness of the most widely used approaches is that they investigate only positive attitudes toward parties. However, there are some examples which clearly show how much considering negative feelings toward unpreferred parties is important when one s aim is to analyse the strength of partisan attachments in a society (Enyedi & Todosijević, 2009; Iyengar et al., 2012; Klingemann & Wattenberg, 1992; Rose & Mishler, 1998b). 3 Robustness checks show that results are highly consistent no matter whether mean or median ratings of winners and losers are used, whether the net differences or the proportions of these two groups are used, whether an index composed by variables of satisfaction and institutional trust or an index of only satisfaction variables is used. 8

9 The method presented here tries to include this theoretically very important feature of partisanship, its scope is to integrate positive and negative biases toward political sides in one variable. Across the differences between the overall political evaluations of the competing camps it shows how strong is the partisan perceptual screen : how much rosier are the evaluations of govering parties voters than that of opposition parties voters. The new indicator offers a completely different picture compared to the existing rankings. In this sense, nowadays the most partisan countries are mainly the Eastern and Southern European ones, like Hungary, Bulgaria, Poland, the Czech Republic, Greece, Spain, Portugal and Cyprus. Partisan polarisation is associated with a lower level of press freedom, with higher level of ideological polarisation, with a more majoritarian political system, and with a less trusting electorate, while there is no relationship between partisan polarisation and the length of democratic experience of a country. All these results and the weak but significant negative relationship between the new method and the traditional one show that they obviously measure completely different features of partisanship. This means that measuring partisanship with the conventional methods may not be very useful for taking inferences on partisan thinking, and partisan bias of voters opinion formation, and based on these results many former statements about the causes and consequences of partisanship seem to be overgeneralizations. Once its content is defined accurately, the conventional measurement method can be useful for other purposes, but as Converse and Pierce stated partisanship has multiple facets, and keeping clear which facet is being measured ( ) is a basic investigator responsibility (1985, p. 143). 9

10 Appendix 1 Countries included in the analysis by ESS round Austria * * * Belgium * * * * * * Bulgaria * * * * Czech Republic * * * * Croatia * Cyprus * * * Denmark * * * * * * Estonia * * * Finland * * * * * * France * * * * * * Germany * * * * * Greece * * * Hungary * * * * * * Italy * Iceland * * Ireland * * * * * Latvia * Lithuania * * Luxembourg * * Netherlands * * * Norway * * * * * * Poland * * * * * * Portugal * * * * * Romania * Slovakia * * * * * Slovenia * * * * * Spain * * * * * * Sweden * * * * * Switzerland * * * * * * United Kingdom * * * * * * 10

11 References Anderson, C. J., & LoTempio, A. J. (2002). Winning, losing and political trust in America. British Journal of Political Science, 32(2), Anderson, C. J., & Tverdova, Y. V. (2001). Winners, losers, and attitudes about government in contemporary democracies. International Political Science Review, 22(4), Blais, A., & Gélineau, F. (2007). Winning, losing and satisfaction with democracy. Political Studies, 55(2), Brunell, T. L., & Buchler, J. (2012). How Does Winning, Losing, and Electoral Competitiveness Affect Voters Attitudes Toward Government? Evidence from Three Western Democracies. Evidence from Three Western Democracies. Burden, B. C., & Klofstad, C. A. (2005). Affect and cognition in party identification. Political Psychology, 26(6), Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American Voter. New York and London: John Wiley & Sons, Inc. Converse, P. E. (1969). Of time and partisan stability. Comparative Political Studies, 2(2), 139. Converse, P. E., & Pierce, R. (1985). Measuring partisanship. Political Methodology, Craig, S. C., Martinez, M. D., Gainous, J., & Kane, J. G. (2006). Winners, losers, and election context: Voter responses to the 2000 presidential election. Political Research Quarterly, 59(4), Curini, L., Jou, W., & Memoli, V. (2012). Satisfaction with democracy and the winner/loser debate: The role of policy preferences and past experience. British Journal of Political Science, 42(2), Dalton, R. J., & Weldon, S. (2007). Partisanship and party system institutionalization. Party Politics, 13(2), Enyedi, Z., & Todosijević, B. (2009). Adversarial politics, civic virtues and partisanship in Eastern and Western Europe. Political Parties and Partisanship: Social Identity and Individual Attitudes, Gerber, A. S., & Huber, G. A. (2009). Partisanship and economic behavior: do partisan differences in economic forecasts predict real economic behavior? Cambridge University Press. Hofstede, G. (1984). Culture s consequences: International differences in work-related values (Vol. 5). sage. Howell, P., & Justwan, F. (2013). Nail-biters and no-contests: The effect of electoral margins on satisfaction with democracy in winners and losers. Electoral Studies, 32(2), Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, not ideology a social identity perspective on polarization. Public Opinion Quarterly, 76(3), Klingemann, H.-D., & Wattenberg, M. P. (1992). Decaying versus developing party systems: A comparison of party images in the United States and West Germany. British Journal of Political Science, 22(2), Körösényi, A. (2013). Political polarization and its consequences on democratic accountability. Corvinus Journal of Sociology and Social Policy, (2), Laakso, M., & Taagepera, R. (1979). The Effective Number of Parties: A Measure with Application to West Europe. Comparative Political Studies, 12(1), 3. Lupu, N. (2013). Party brands and partisanship: Theory with evidence from a survey experiment in Argentina. American Journal of Political Science, 57(1),

12 Miller, A. H., & Klobucar, T. F. (2000). The development of party identification in post- Soviet societies. American Journal of Political Science, Palonen, E. (2009). Political polarisation and populism in contemporary Hungary. Parliamentary Affairs, 62(2), Rose, R., & Mishler, W. (1998a). Negative and positive party identification in postcommunist countries. Electoral Studies, 17(2), Rose, R., & Mishler, W. (1998b). Negative and positive party identification in postcommunist countries. Electoral Studies, 17(2), Samuels, D. (2006). Sources of mass partisanship in Brazil. Latin American Politics and Society, 48(2), Singh, S., Karakoç, E., & Blais, A. (2012). Differentiating winners: How elections affect satisfaction with democracy. Electoral Studies, 31(1), Singh, S., Lago, I., & Blais, A. (2011). Winning and Competitiveness as Determinants of Political Support. Social Science Quarterly, 92(3), Tardos, R., & Angelusz, R. (2009). A kapcsolathálózati szemlélet a társadalom-és politikatudományban. Politikatudományi Szemle, (2), Tóka, G. (2005). A magyarországi politikai tagoltság nemzetközi összehasonlításban. In R. Angelusz & R. Tardos (Eds.), Törések, hálók, hidak (pp ). Data Armingeon, Klaus, Christian Isler, Laura Knöpfel, David Weisstanner and Sarah Engler Comparative Political Data Set Bern: Institute of Political Science, University of Berne. Armingeon, Klaus, Christian Isler, Laura Knöpfel and David Weisstanner Supplement to the Comparative Political Data Set Government Composition Bern: Institute of Political Science, University of Berne. ESS Round 6: European Social Survey Round 6 Data (2012). Data file edition 2.2. NSD - Norwegian Centre for Research Data, Norway Data Archive and distributor of ESS data for ESS ERIC. ESS Round 5: European Social Survey Round 5 Data (2010). Data file edition 3.2. NSD - Norwegian Centre for Research Data, Norway Data Archive and distributor of ESS data for ESS ERIC. ESS Round 4: European Social Survey Round 4 Data (2008). Data file edition 4.3. NSD - Norwegian Centre for Research Data, Norway Data Archive and distributor of ESS data for ESS ERIC. ESS Round 3: European Social Survey Round 3 Data (2006). Data file edition 3.5. NSD - Norwegian Centre for Research Data, Norway Data Archive and distributor of ESS data for ESS ERIC. ESS Round 2: European Social Survey Round 2 Data (2004). Data file edition 3.4. NSD - Norwegian Centre for Research Data, Norway Data Archive and distributor of ESS data for ESS ERIC. ESS Round 1: European Social Survey Round 1 Data (2002). Data file edition 6.4. NSD - Norwegian Centre for Research Data, Norway Data Archive and distributor of ESS data for ESS ERIC. Polity IV Project: Political Regime Characteristics and Transitions,